rec10/trunk/src/classify.py

   1 #!/usr/bin/python
   2 # coding: UTF-8
   3 # Rec10 TS Recording Tools
   4 # Copyright (C) 2009 Yukikaze
   5
   6 import glob
   7 import n_gram
   8 import os
   9 import os.path
  10 import re
  11 import time
  12 import optparse
  13 import sys
  14 import shutil
  15
  16 import configreader
  17 import recdblist
  18 recordedpath=unicode(configreader.getpath("recorded"),'utf-8')
  19 recpath=unicode(configreader.getpath("recpath"),'utf-8')
  20 def search_b25(path):
  21     """
  22     録画フォルダを検索し、各QUEの実行状況を表示する
  23     """
  24     b25list = glob.glob(path + "/*.b25")
  25     for b25f in b25list:
  26         ##b25f is title.ts.b25  avi is title.avi
  27         dir = os.path.split(b25f)[0]
  28         title = os.path.split(b25f)[1]
  29         title = title.replace(".ts.b25", "")
  30         avipath = os.path.join(dir, title + ".avi")
  31         mp4path = os.path.join(dir, title + ".mp4")
  32         tspath = os.path.join(dir, title + ".ts")
  33         b25f = b25f.replace(".ts.b25", ".avi")
  34         mode = "ts"
  35         if os.path.isfile(tspath):##tsファイルが存在している
  36             dtime = time.time()-os.path.getmtime(tspath)
  37             dtime = int(dtime)
  38             if dtime > 120:
  39                 if os.path.getsize(tspath) > 1 * 1000 * 1000:##最終更新から22分以上経過かつ1MB以上
  40                     mode = "avi"
  41                 else:
  42                     mode = "tsmiss"
  43             else:
  44                 mode = "ts"
  45         if os.path.isfile(mp4path):##tsファイルが存在している
  46             dtime = time.time()-os.path.getmtime(mkvpath)
  47             dtime = int(dtime)
  48             if dtime > 120:
  49                 if os.path.getsize(avipath) > 1 * 1000 * 1000:##最終更新から22分以上経過かつ1MB以上
  50                     mode = "fin"
  51                 else:
  52                     mode = "avimiss"
  53             else:
  54                 mode = "avi"
  55         recdblist.printutf8(title + ":" + mode)
  56 def search_avi(temppath,recpath):
  57     """
  58     録画一時フォルダ内のaviファイルを検索
  59     """
  60     avilist = glob.glob(temppath + "/*.avi")
  61     ret=[]
  62     for avif in avilist:
  63         ##b25f is title.ts.b25  avi is title.avi
  64         dir = os.path.split(avif)[0]
  65         title = os.path.split(avif)[1]
  66         title = title.replace(".avi", "")
  67         avipath = os.path.join(dir, title + ".avi")
  68         dtime = time.time()-os.path.getmtime(avipath)
  69         dtime = int(dtime)
  70         if dtime > 300:
  71             if veryfySize(avipath):
  72                 ret.append(title)
  73     return ret
  74 def search_mkv(temppath,recpath):
  75     """
  76     録画一時フォルダ内mkvファイルを検索
  77     """
  78     avilist = glob.glob(temppath + "/*.mkv")
  79     ret=[]
  80     for avif in avilist:
  81         ##b25f is title.ts.b25  avi is title.avi
  82         dir = os.path.split(avif)[0]
  83         title = os.path.split(avif)[1]
  84         title = title.replace(".mkv", "")
  85         avipath = os.path.join(dir, title + ".mkv")
  86         dtime = time.time()-os.path.getmtime(avipath)
  87         dtime = int(dtime)
  88         if dtime > 300:
  89             if veryfySize(avipath):
  90                 ret.append(title)
  91     return ret
  92 def search_mp4(temppath,recpath):
  93     """
  94     録画一時フォルダ内mp4ファイルを検索
  95     """
  96     avilist = glob.glob(temppath + "/*.mp4")
  97     ret=[]
  98     for avif in avilist:
  99         ##b25f is title.ts.b25  avi is title.avi
 100         dir = os.path.split(avif)[0]
 101         title = os.path.split(avif)[1]
 102         title = title.replace(".mp4", "")
 103         avipath = os.path.join(dir, title + ".mp4")
 104         dtime = time.time()-os.path.getmtime(avipath)
 105         dtime = int(dtime)
 106         if dtime > 300:
 107             if veryfySize(avipath):
 108                 ret.append(title)
 109     return ret
 110 def veryfySize(path):
 111     #vsize=[297,497,596,1196]#SD 30m 1h 1.5h 2h
 112     vsize=[245,275,295,591,830]
 113     vsize=vsize+[325,449,560,590,602,690,805,860,1014,1138,1237]
 114     #vsize=vsize+[458,916]#HD 30m 1h
 115     #vsize=vsize+[381,895,447]
 116     ret = 0
 117     for size in vsize:
 118         if os.path.getsize(path)>(size-10)*1024*1024 and os.path.getsize(path)<(size+10)*1024*1024:
 119             ret=1
 120     return ret
 121 def searchFolder(title,path):
 122     """
 123     titleにマッチするフォルダを探し出す。
 124     """
 125     folderpath=os.listdir(path)
 126     lfpath=[]
 127     ngram=[]
 128     for ft in folderpath:
 129         fullpath=os.path.join(path.encode('utf-8'), ft)
 130         if os.path.isdir(fullpath):
 131             lfpath.append(fullpath)
 132             ftt=os.listdir(fullpath)
 133             if len(ftt)>0:
 134                 for ft2 in ftt:
 135                     try:
 136                         folderpath.append(os.path.join(fullpath, ft2))
 137                     except Exception, inst:
 138                         #print type(inst)
 139                         #print inst
 140                         ""
 141         else:
 142             lfpath.append(fullpath)
 143     for dirp in lfpath:
 144         cmpp=""
 145         appp=""
 146         ntitle=title
 147         if os.path.isdir(dirp):
 148             cmpp=os.path.dirname(dirp)
 149             appp=dirp
 150         else:
 151             cmpp=os.path.basename(dirp)
 152             appp=os.path.dirname(dirp)
 153         rT=re.compile("(.+)_(\d*)\Z")
 154         tT=rT.match(title)
 155         tT2=rT.match(cmpp)
 156         if tT:
 157             ntitle=tT.group(0)
 158         if tT2:
 159             cmpp=cmpp.group(0)
 160         p=n_gram.trigram(ntitle,cmpp)
 161         if p>0:
 162             ngram.append((p,appp))
 163     ngram=list(set(ngram))
 164     ngram.sort()
 165     ngram.reverse()
 166     if len(ngram)>0:
 167         #recdblist.printutf8(title + ngram[0][1] + " : "+str(ngram[0][0]))
 168         if ngram[0][0]>400:
 169             return ngram[0][1]
 170         else:
 171             return ""
 172     else:
 173         return ""
 174 def execMove(title,temppath,recpath):
 175     srcpath=os.path.join(temppath,title+".mp4")
 176     #desttitle=destName(title, temppath, recpath)
 177
 178     sf=searchFolder(title, recpath)
 179     if sf!="":
 180         destpath=os.path.join(sf,destNameMP4(title, temppath, sf)+".mp4")
 181         #os.path.join(sf,desttitle+".mkv")
 182         #recdblist.printutf8(srcpath)
 183         #recdblist.printutf8(destpath)
 184         recdblist.printutf8("moving now..")
 185         recdblist.printutf8(srcpath+" : "+destpath)
 186         shutil.move(srcpath, destpath)
 187         #shutil.copy(srcpath, destpath)
 188         delpath=[os.path.join(temppath,title+".ts")]
 189         delpath.append(os.path.join(temppath,title+".avi"))
 190         delpath.append(os.path.join(temppath,title+".ts.b25"))
 191         delpath.append(os.path.join(temppath,title+".ts.tsmix"))
 192         delpath.append(os.path.join(temppath,title+".ts.log"))
 193         delpath.append(os.path.join(temppath,title+".sa.avi"))
 194         delpath.append(os.path.join(temppath,title+".sa.avi.log"))
 195         delpath.append(os.path.join(temppath,title+".log"))
 196         for dp in delpath:
 197             try:
 198                 os.remove(dp)
 199                 ""
 200             except:
 201                 ""
 202 def destName(title,temppath,recpath):
 203     """
 204     titleから移動先に作るべきファイル名を作り出す。
 205     """
 206     dstpath=os.path.join(recpath,title+".avi")
 207     srcpath=os.path.join(temppath,title+".avi")
 208     if os.path.exists(dstpath):
 209         gmtime=time.gmtime(os.path.getmtime(srcpath))
 210         times=time.strftime("%y%m%d",gmtime)
 211         title=title+"_"+times
 212     return title
 213 def destNameMKV(title,temppath,recpath):
 214     """
 215     titleから移動先に作るべきファイル名を作り出す。
 216     """
 217     dstpath=os.path.join(recpath,title+".mkv")
 218     srcpath=os.path.join(temppath,title+".mkv")
 219     if os.path.exists(dstpath):
 220         gmtime=time.gmtime(os.path.getmtime(srcpath))
 221         times=time.strftime("%y%m%d",gmtime)
 222         title=title+"_"+times
 223     return title
 224 def destNameMP4(title,temppath,recpath):
 225     """
 226     titleから移動先に作るべきファイル名を作り出す。
 227     """
 228     dstpath=os.path.join(recpath,title+".mp4")
 229     srcpath=os.path.join(temppath,title+".mp4")
 230     if os.path.exists(dstpath):
 231         gmtime=time.gmtime(os.path.getmtime(srcpath))
 232         times=time.strftime("%y%m%d",gmtime)
 233         title=title+"_"+times
 234     return title
 235 def detName(path,title):
 236     """
 237     type A ---title#<number>
 238     type B ---title#<number>subtitle
 239     type C ---title subtitle
 240     type D ---title(without number)
 241     """
 242     #if re.match("#\d\s[0,10]\z|#[0-9]\s[0,10]\z", title)
 243     #    recdblist.printutf8("typeA")
 244     #elif re.match("#\d\s[0,10].|#[0-9]\s[0,10].", title)
 245     #    recdblist.printutf8("typeB")
 246
 247 def detNameType(title):
 248     """
 249     type A ---title#<number>
 250     type B ---title#<number>subtitle
 251     type C ---title subtitle
 252     type D ---title(without number)
 253     """
 254     recdblist.printutf8(title)
 255     #rA=re.compile(".+(?P<title>)#\d(?P<num>)\s[0,10]\z")
 256     rA=re.compile("(.+)#(\d*)\s*\Z")
 257     tA=rA.match(title)
 258     rB=re.compile("(.+)#(\d*)\s*(\D*)")
 259     tB=rB.match(title)
 260     if tA:
 261         recdblist.printutf8("typeA")
 262         recdblist.printutf8("title="+tA.group(1))
 263         recdblist.printutf8("num="+tA.group(2))
 264     elif tB:
 265         recdblist.printutf8("typeB")
 266         recdblist.printutf8("title="+tB.group(1))
 267         recdblist.printutf8("num="+tB.group(2))
 268         recdblist.printutf8("subtitle="+tB.group(3))
 269 if __name__ == "__main__":
 270     usage="usage: %prog read -h"
 271     version="%prog 0.7.0beta1"
 272     parser=optparse.OptionParser(usage=usage,version=version)
 273     parser.add_option("-s","--search",action="store",type="string",dest="ltitle",default="",metavar="TITLE",help="test to search where the title should be contained(test for -e)")
 274     parser.add_option("-A","--Auto",action="store_true",dest="auto",default=False,help="auto classifying mode(not a test)")
 275     parser.add_option("-e","--exec",action="store",type="string",dest="etitle",default="",metavar="TITLE",help="exec move(not a test)")
 276     parser.add_option("-l","--list",action="store_true",dest="list",default=False,help="File listing mode(test for -a)")
 277     (opts,args)=parser.parse_args(sys.argv)
 278     if opts.ltitle!="":
 279         #print opts.ltitle
 280         ltitle=unicode(opts.ltitle,'utf-8')
 281         sf=searchFolder(ltitle, recordedpath)
 282         recdblist.printutf8(os.path.join(sf, destNameMP4(ltitle,recpath,sf)+".mp4"))
 283         #print os.path.join(sf, destNameMP4(ltitle,recpath,sf)+".mp4")
 284     elif opts.auto:
 285         sa=search_mp4(recpath, recordedpath)
 286         for t in sa:
 287             recdblist.printutf8(u"自動推測実行中")
 288             sf=searchFolder(t,recordedpath)
 289             if sf!="":
 290                 recdblist.printutf8(u"移動先")
 291                 recdblist.printutf8(t+" : "+os.path.join(sf, destNameMP4(t, recpath,sf)+".mp4"))
 292                 recdblist.printutf8(u"実行中")
 293                 execMove(t, recpath, recordedpath)
 294             else:
 295                 recdblist.printutf8(t+" can't find matching folder.")
 296     elif opts.list:
 297         sa=search_mp4(recpath, recordedpath)
 298         for t in sa:
 299             sf=searchFolder(t,recordedpath)
 300             if sf!="":
 301                 recdblist.printutf8(t+u" : "+os.path.join(sf, destNameMP4(t, recpath,sf)+".mp4"))
 302     elif opts.etitle!="":
 303         etitle=unicode(opts.etitle,'utf-8')
 304         execMove(etitle,recpath, recordedpath)
 305